The field guide
Read the finding. Follow the mechanism. Ask what comes next.
A guide to what studies establish, what mechanisms make plausible, and which observations would move an explanation forward.
Evidence does two jobs: it answers a question and helps reveal the next one. A biochemical mechanism, a microbial pattern, or a change in metal handling can provide a strong reason to investigate a new connection. Read the finding, the proposed explanation, and its predictions separately so you can see both the contribution and the next opportunity.
1. Check the publication record
- Preprint
- A public manuscript whose posting does not establish journal peer review. Check its version and whether a later journal article exists.
- Published research article
- Check the publisher's article type, dates, corrections, and methods. Publication does not guarantee that a conclusion will hold.
- Review or commentary on this site
- An interpretation of a cited source. It is distinct from that paper and from a journal's peer-review process. Follow the primary source before quoting a result.
- DOI
- A persistent identifier for a record. It does not by itself establish peer review, study quality, or independence.
Our editorial standards explain the site's publication labels.
2. Match the claim to the design
| Design | Useful for | Questions to keep open |
|---|---|---|
| Biochemical or in-vitro experiment | Testing a defined interaction in controlled conditions. | Are the chemical form, matrix, and exposure relevant to the proposed biological setting? |
| Animal study | Studying a whole-organism response and candidate mechanisms. | How do species, dose, diet, housing, and microbiome affect transfer to humans? |
| Observational human study | Describing associations, timing, variation, and real-world exposures. | Could confounding, reverse causation, selection, or measurement error explain the association? |
| Randomized trial | Estimating an intervention's effect in the studied conditions. | Were allocation, adherence, missing data, outcomes, and follow-up handled adequately? |
| Systematic review or meta-analysis | Bringing a defined body of evidence together. | Was the search reproducible? Were study quality, missing results, and differences between studies addressed? |
| Mechanistic synthesis or hypothesis | Explaining connections, identifying overlooked variables, and deriving predictions that can guide new work. | What findings ground the mechanism, which steps are inferred, and what result would distinguish the proposed explanation from alternatives? |
These prompts are not formal quality scores. Cochrane emphasizes evaluating a design's features and potential biases, rather than relying only on its label. Methods guidance ↗
3. Build the bridge from mechanism to hypothesis
Begin with what the work actually established. Then state the new connection it motivates and why the mechanism would operate in the proposed setting. A useful hypothesis produces a prediction that distinguishes it from at least one alternative explanation.
- Grounding: Which observation or established mechanism makes this idea plausible?
- Bridge: Which additional step connects that finding to the proposed microbial, host, or disease outcome?
- Prediction: What result would be expected if this explanation is right, and what different result would favor another explanation?
An unmeasured step becomes a research question. It is neither proof of the connection nor a reason to abandon it. A null result becomes informative when the study was capable of observing the effect the hypothesis predicts. Follow the measurement and its uncertainty before deciding what the result changes.
4. Follow the measurement to the claim
- Metal: total concentration, chemical species, bound fraction, location, or an exposure proxy?
- Microbe: relative abundance, absolute abundance, genes, gene expression, metabolites, or measured function?
- Host: a biomarker, symptoms, a clinical outcome, or an inferred pathway?
- Time: did exposure precede the microbial change and outcome, or were they measured together?
- Comparison: which population, matrix, dose, duration, and comparator does the estimate describe?
When measurements change together, ask which directions of influence could explain the pattern. A proposed microbial mediator gives researchers an intervening pathway to examine; a shared cause or a host-driven change may explain the same association differently. Identifying the measurements that separate these explanations is a productive result of mechanistic reasoning. Read about confounding and other biases ↗
5. Compare apparently conflicting results
The library includes an infant iron-fortification trial in Kenya and a short supplementation trial in Australian women. They study different populations and conditions. Record those differences before treating the results as a contradiction or as a general rule.
Check effect sizes and uncertainty, not only a significance threshold. A nonsignificant result is not automatically evidence of equivalence. Across a body of research, consider risk of bias, consistency, relevance, precision, and missing evidence. Cochrane guidance on certainty ↗
A reading note that leads somewhere
- The study asked…
- It measured… in… compared with…
- It found… with this uncertainty…
- This matters because the mechanism…
- That suggests the hypothesis…; another explanation is…
- The next observation that would distinguish them is…
Use this structure when suggesting a paper, proposing a research direction, or contributing a correction. A useful contribution can establish a finding, sharpen an explanation, or identify the next decisive question.